Kelly Marie Angel:

Could it be that the monitors on the affected ring are on a different phase to the main computer in the server room? And when certain operations take place something in the leads is being frazzled. Just a thought. Sparky Dave looks like we are thinking along the same sort of lines also we posted within a few seconds of each other thats why I didn't see your post till after I posted mine

Hi Kelly, the server is on L2 the same as the ring so unless the UPS is kicking in and doing weird stuff which I don't think is the case then it's got to be something else

mapj1:

Please do not be put off by DZ - he comes in at the end of a stressful day sometimes and it shows - it is not just you  - he is like that with the rest of us too.

(now I shall have to duck...)


Right, well you have been a busy bee. Thanks for all that hard work. Digital persist is jolly useful on a scope isn't it?


What follows may be complete hogwash, it is as they say 'your party'. However, here is how it feels to me, having read it.

I think we should assume for now probably only really borking leads with electronics in, and the IT folk who had hold of it before you are either not remembering it right when they tell you, or they were special HDMI leads that had frame rate converters in or something, but in any case were not simple passive.

Display port - ah yes a right pain in the interface being 3V logic, when DVI and HDMI are 5V. Expensive if the level shifters are omitted in the cheap far eastern model


I had not clocked immediately that the clean and dirty rings were 2 phases on the same DB, so two lives and a common neutral.


That opens the door to a fun test. IF they can be swapped at the breaker, does the problem also swap to the other side of the room, or safer, if moved both to the clean phase, does the problem go away ?

I.e. have you really got one dirty phase coming into the board? I suspect you have something like that, but it is hard to prove. If the sub-main feeding the DB is SWA or waveform, then it forms a kind of coax transmission line, so spikes can be conducted in from quite a distance away without much attenuation.


The short duration transients may be partly radiated - faster then about a microsecond risetime, (well for wiring of tens of metres long) and it is often as easy  or easier for the energy to get off the wire and back on again further down in a radio and antennas or even a loosely guided wave sort of way rather than in a purely conducted way.

I suspect your remote light switch may be re-creating a version of  the famous experiment of Heinrich Herz    but with the scope serving as a rather more sensitive detector than Herz's original spark gap.


So, what to do? Well first stop plugging in and flashing lights ;-)

More seriously, it seems there are fast spikes coming in on the second phase, so some sort of filtering at the DB may be worth installing, of the kind that puts quite some capacitance from L to E and N to E, to absorb any fast rising edges inbound, Verify that the  N voltage is not moving relative to the lives, by looking at the NE waveform on the scope at the DB, or as near as you dare get to it,  You may care to beef up the earthing. Threading some ferrite on the inbound cables will have no ill effect on the 50Hz, but will raise the series impedance to provide opposition to microsecond scale stuff. (example)


There is one other possibility to be eliminated - and that is one of the loads on the dirty ring is acting as the impulse generator, perhaps with partial 'crackly'  break down of a varistor or internal filter capacitance inside the PSU in one PC or the other when it warms up. Or a scratchy current carrying contact in a plug, socket or mains lead. This would be the suspect if the swap phase test brings the problem with it..


I assume you would have mentioned other details like metal suspended ceilings or floors, but if there are then these are possible re-radiators.




Mike.

Edit do not be upset by the phase inversion - the scope will trigger on both positive and negative edges, and a sine wave with some zero-crossing distortion that includes an occasional 'twitch' or reversal of direction may cause it to trigger on a short up ring during a down edge, so it appears to fire on  the wrong edge occasionally, and the persist on the scope makes it look really weird.

Traces with fuzz riding on them like the chairs on a ski-lift are hard for the scope to get a grip on, even a modern digital one.


Morning,


I'm going to digest this and try to do some more investigation throughout the day so I will update you on any findings regarding your suggestions and thank you, it is appreciated as i'm starting to feel under the microscope some what on this drawn out affair!


At the risk of giving too much info prior to trying your suggestions I just want to add what I discovered late on yesterday as I feel it may change a few theories.


So the "clean" ring on the other side of the room is where I described doing the exact same test and only found small (5V) spikes on the monitor casing (Green line on scope)... well I decided to try the same thing again on another randomly chosen monitor on the "clean" ring again and I found that the spikes were more prevalent when switching the lamp (which is now being switched on a completely different phase ring), just not on the scale of the ring causing all the drama. I then asked the office staff once again, " are you sure you have never noticed any monitor issues like what the other side have been getting" but somebody piped up and said that they don't get issues generally but he has replaced faulty leads once ages ago as his monitor did turn off! So that's once amongst a group of 12 people on the L1 ring (clean) and about say 200 times total on the L2 ring but once isn't not at all!


Something that I thought about this morning but I might be barking up the wrong tree as it doesn't explain why one side is worse than the other but all of the PC stations are Covid safe, with perspex dividers and they are all made up of steel units and the monitors are all mounted on steel shrouds.... could this be forming some kind of static build up or creating some transient voltages some how? All new (ish) carpets ?‍♂️ but this wouldn't explain the first couple of months with no issues


I have taken some monitors off the steel shrouds so it's a case of waiting and seeing over the next few days on that one.


Another thought, these faulty leads (none passive) could be from the same batch (still investigating) even though they all vary in type in which case, there's a global issue regarding supply of semi conductors so i'm wondering if the supplier has had to find a dodgy alternative and that is highlighted by the "Dirty" ring (sounds wrong) but maybe what i'm seeing on the scope is not that out of the ordinary like has been previously suggested? I mean, I still don't understand what i'm seeing at all ? but still, that's a theory at least. I'll let you know if I can get any clarity on that.

 

davezawadi (David Stone):

Dave, sorry if I sounded a bit short, it comes with the territory as Mike says.

Displays go blank for a short period for a number of reasons, it possibly could be EMC, but as the computer software itself is not crashing, it is probably not affecting the PC operation. Changing the Graphics settings does sometimes blank the screen for a second or two, as does plugging in a second display, or switching between two programs that REQUIRE specific graphics settings, such as certain CAD and accounting packages, of a slightly older vintage. Fully compliant Windows 10 programs should not do this (I assume all the computers are running the same OS etc. and are used for similar purposes if not everything gets even more complex!). I cannot see a way where one phase is much worse than another, but there is an easy test, simply swap over the outputs of the 2 breakers and see if the fault follows, and preferably the waveforms too. As Mike says, this possibly could be a connected appliance to the bad ring, an SMPS that misbehaves in a strange way if the supply changes slightly, faulty filter components (self-healing capacitors can cause very strange things to happen when wearing out). Swapping breakers then will give you an idea of where to start elimination, but it will be time-consuming.


Because you don't know if the leads are actually faulty (and if the image comes back, they are very unlikely to be) this also might be a red herring. There is the possibility this is due to a virus on some of the computers but then the spikes etc. are very curious as they cannot be connected to a program, except in very tenuous ways. It is also possible that remote desktop could cause the video breaks, again very strange.


The best next step is definitely to swap the supply over.


Kind regards

David

No problem, I know how it is..


Sorry, just to clarify, there's now a question mark over there being an issue with any HDMI to HDMI leads but the other types are 100% faulty. They start off by constantly switching the monitor off and on but then some but not all, end up not switching back on again after a while of flickering. These have then had to have their leads replaced and all the faulty one's have been sent to the IT dept down in Bude for them to investigate that they are faulty before they send more up (to leicester) so a very long winded affair! They have taken leads that have stopped working and tried them on a machine that has had no problems and they still don't work.


The IT guys have also ruled out any driver updates / windows 10 issues also.


I'm a couple of steps away from trying your suggestion but I will keep you posted! Fingers crossed!

Roger Bryant:

Going back a few posts you said:

'There's no potential difference between the earths on either ring. 00.01V's and 4 ohms between the two using the wandering lead method'.

4 ohms between the earths on the rings sounds a lot to me. How did you measure it? Was the system live or dead at the time?

 

You make a good point, that's made me go back and double check but this time I took the socket fronts off . I didn't before as it is all still live and didn't want to risk disturbing any of the CAD teams work so I just went from the screw heads on the socket fronts using a low power multi meter. This time it has read 1.5 ohms and that is with the wandering lead resistance deducted.

Sparky Dave:

Roger Bryant:

Going back a few posts you said:

'There's no potential difference between the earths on either ring. 00.01V's and 4 ohms between the two using the wandering lead method'.

4 ohms between the earths on the rings sounds a lot to me. How did you measure it? Was the system live or dead at the time?

 

You make a good point, that's made me go back and double check but this time I took the socket fronts off . I didn't before as it is all still live and didn't want to risk disturbing any of the CAD teams work so I just went from the screw heads on the socket fronts using a low power multi meter. This time it has read 1.5 ohms and that is with the wandering lead resistance deducted.

and that's live still